Re: [PATCH 03/17] blktrans: track open and close calls.

[Maxim Levitsky <maximlevitsky@gmail.com>]

On Mon, 2010-02-01 at 08:46 +0100, Ricard Wanderlof wrote: 
> On Sat, 30 Jan 2010, Maxim Levitsky wrote:
> 
> > This patch adds tracking for open and close calls.
> > Now trans ->open and ->release are never called twise in a row
> > ->release is also called once before mtd device disappers
> > ...
> 
> > --- a/drivers/mtd/mtd_blkdevs.c
> > +++ b/drivers/mtd/mtd_blkdevs.c
> > @@ -128,6 +128,9 @@ static int blktrans_open(struct block_device *bdev, fmode_t mode)
> > 	if (!get_mtd_device(NULL, dev->mtd->index))
> > 		goto out;
> >
> > +	if (dev->open++)
> > +		goto out;
> > +
> > 	if (!try_module_get(tr->owner))
> > 		goto out_tr;
> >
> > @@ -153,6 +156,10 @@ static int blktrans_release(struct gendisk *disk, fmode_t mode)
> > 	struct mtd_blktrans_ops *tr = dev->tr;
> > 	int ret = 0;
> >
> > +	dev->open--;
> > +	if (dev->open)
> > +		return 0;
> > +
> 
> Just a very minor quibble: if you use if (dev->open++) in one function, 
> why not use if (--dev->open) in the other? Or separate the 
> increment/decrement from the test, depending on you style preferences.
No problem, this just got lost in all the noise.
I tried all kinds of approaches to make hotplug reliable, before finally
making it be so. (It is really stable here, I didn't have any hotplug
related issues for very long time)

In fact I feel that my xd card is more stable that sd one...
(I recently found a bug in sdhci driver....)

I would be very happy to receive a full review of the patches, and get
it in the kernel.

Best regards,
Maxim Levitsky